Open-Meteo Solar Forecast

This project is based on the Forecast.Solar integration but makes use of the Open-Meteo API to generate the forecasts. The integration could be installed from GitHub - rany2/ha-open-meteo-solar-forecast: Home Assistant Open-Meteo Solar Forecast Integration. You could install it via HACS by adding it as a “Custom repository” or manually.

For me, the results are better than Forecast.Solar (at least their Free tier) and on-par (if not slightly better) than Solcast. Of course, the primary advantage is that you get data at 15-min intervals (though some parts of the world make use of interpolation) and that you could poll the API more frequently without worrying about reaching a rate-limit. This means that the power sensors it provides are more precise. I’ve been using it personally for a few weeks now after Solcast broke my scrapper script :')


wanted to try but got an error during setup

Logger: homeassistant.util.package
Source: util/
First occurred: 18:50:55 (3 occurrences)
Last logged: 18:51:11

Unable to install package open_meteo_solar_forecast==0.1.5: ERROR: Could not find a version that satisfies the requirement open_meteo_solar_forecast==0.1.5 (from versions: 0.1.1, 0.1.2, 0.1.3, 0.1.4) ERROR: No matching distribution found for open_meteo_solar_forecast==0.1.5

Uploaded, now sorry! You might need to restart twice due to pypi caching.

P.S. @elRadix please update, I made a calculation mistake. This code was ported from Go to Python just a few hours ago and I made an error. It actually looks good now. I was previously using a Forecast.Solar compatible HTTP service so I could just change the host and call it a day but it’s now it’s own thing :slight_smile:

1 Like

OK, it’s working now
I’ll update and follow pv forecast progress.

1 Like

I’ve added some cloud cover compensation code in v0.1.1 as it seems like Open-Meteo’s cloud cover code isn’t enough.

1 Like

Like the look of this. Much more accurate for me than, and less drama :smiley: than the solcast integration…
Installed with two instances for two arrays, and will monitor! Thanks!

This is with this configured for 2 arrays at a 95% efficiency, based on my DC panels size, not inverter size:

And this is Solcast for the same 2 arrays:

I installed it yesterday but the estimates differ vastly from the SolCast ones (that have proven pretty accurate so far). I have no clue why. Settings are the same as for SolCast.

I also noticed that I have several disabled sensors - why?

Hey all, I’m surprised there’s a ton of sudden activity here. I generally prefer if you make GitHub issues over using this forum but I’ll try my best to reply to everyone here.


I installed it yesterday but the estimates differ vastly from the SolCast ones (that have proven pretty accurate so far). I have no clue why. Settings are the same as for SolCast.

Unfortunately without more information I can’t really help you, could you provide me with your solcast config and open-meteo solar config? Also when you say that it differs vastly, is it overestimating or underestimating?

I also noticed that I have several disabled sensors - why?

Some sensors are disabled by default but you could enable them, the reason is so that your recorder database doesn’t explode :slight_smile:

Thanks a lot for the offer but I don’t want to share my location.

It is underestimating a lot - SolCast for today says nearly 60 kWh. Yours only 18.6 kWh…

@andreas-bulling Do you have the azimuth set properly? The azimuth setting for this integration is [0,360] (0=North not South), so if you took the measurements in [-180,180] (0=North) and have a negative value you need to add 360 to make it positive.

1 Like

@andreas-bulling Please try version 0.1.2 of the integration. I removed the cloud cover compensation code. I have a feeling that Open-Meteo’s cloud cover code works well over certain regions only and I’ll have to be more selective about applying it.

1 Like

I had the same issue (reporting way too low value for daily forcast). The update seems to be great for me. Todays forecast went up from 3.2kWh to 6.6kWh. Solcast is reporting 8kWh. I have generated 5.8kWh with not much daylight left today … so the 6.6 seems very realistic to me. Thanks :slight_smile:

That’s great, thanks a lot for reporting issues! P.S. for the efficiency factor, I recommend you start with 0.93 as a baseline to account for DC wire loss.

FYI I’ve applied v0.1.2 and the forecast for tomorrow now looks much better. I’ve got some cloud forecast and the energy predicted for tomorrow has jumped from 3.6 (which was too low) to 9.3 (which looks about right. Thanks for setting up this repository.

I was just about to write that it might be the azimuth angle…

Our panels are inclined 30 degrees with an azimuth of -160 degrees (south, SolCast setting).

This translates into 160 degrees in your integration, correct?

Trying to test 0.1.2 but it seems I exceeded the rate limit…

That’s definitely wrong, you just add 360 to make it nonnegative. So it should be 200 in this case…

How do you know that? The rate limit is 10,000 per IP per day, so you should never realistically hit it.